Merge pull request #67 from elliot-100/refine_get_public

Refine `get_public_Info()`
elliot-100 · Nov 28, 2023 · fc4a6c2 · fc4a6c2
2 parents 0cd745f + ad4ac40
commit fc4a6c2
Show file tree

Hide file tree

Showing 2 changed files with 49 additions and 7 deletions.
diff --git a/britishcycling_clubs/__init__.py b/britishcycling_clubs/__init__.py
@@ -141,19 +141,21 @@ def get_public_club_info(club_id: str) -> PublicClubInfo:
     -------
     PublicClubInfo
     """
-    profile_page = requests.get(
-        f"{PROFILE_BASE_URL}{club_id}/",
-        timeout=REQUESTS_TIMEOUT,
-    )
-    profile_soup = BeautifulSoup(profile_page.content, "html.parser")
+    profile_url = f"{PROFILE_BASE_URL}{club_id}/"
+    r = requests.get(profile_url, timeout=REQUESTS_TIMEOUT)
+    r.raise_for_status()
+    if r.url != profile_url:
+        error_message = f"Redirected to unexpected URL {r.url}. Is `club_id` valid?"
+        raise ValueError(error_message)
+    profile_soup = BeautifulSoup(r.content, "html.parser")
     return {
         "club_name": _get_club_name_from_profile(profile_soup),
         "total_members": _get_total_members_from_profile(profile_soup),
     }
 
 
 def _get_club_name_from_profile(soup: BeautifulSoup) -> str:
-    """Return the club's name from BeautifulSoup object."""
+    """Return the club's name from profile page soup."""
     club_name_h1 = soup.find("h1", class_="article__header__title-body__text")
 
     # For type-checking purposes: ensures unambiguous type is passed
@@ -168,7 +170,7 @@ def _get_club_name_from_profile(soup: BeautifulSoup) -> str:
 
 
 def _get_total_members_from_profile(soup: BeautifulSoup) -> int:
-    """Return the club's total members count from BeautifulSoup object."""
+    """Return the club's total members count from profile page soup."""
     about_div = soup.find("div", id="about")
 
     # TypeError is raised if page other than a club profile page is returned

diff --git a/tests/test_functions.py b/tests/test_functions.py
@@ -0,0 +1,40 @@
+"""Tests for functions."""
+
+from bs4 import BeautifulSoup
+
+from britishcycling_clubs import (
+    _get_club_name_from_profile,
+    _get_total_members_from_profile,
+)
+
+# Partial extract from actual page
+PROFILE_PAGE_EXTRACT = """
+        <html>
+        
+        <h1 class="article__header__title-body__text">My Cycling Club</h1>
+    
+        <div class="tabs__content" id="about">
+        <h2>Club Information</h2>
+        <div class="grid-tablet-wrap grid-tablet-wide-wrap grid-desktop-wrap">
+        <div class="grid-tablet-col-full grid-tablet-wide-col-1-2 grid-desktop-col-1-2">
+        <p>
+        <b>Club type:</b> Club Commercial<br/>
+        <b>Affiliated: </b>31/12/2023<br/>
+        <b>Affilation status:</b> Active </p>
+        <p><b>Total club members:</b> 42<br/></p>
+        </div>
+
+        </html>
+"""
+
+
+def test__get_club_name_from_profile__happy_path() -> None:
+    """Test that club name is returned from 'profile' soup."""
+    profile_soup = BeautifulSoup(PROFILE_PAGE_EXTRACT)
+    assert _get_club_name_from_profile(profile_soup) == "My Cycling Club"
+
+
+def test__get_total_members_from_profile__happy_path() -> None:
+    """Test that total members count is returned from 'profile' soup."""
+    profile_soup = BeautifulSoup(PROFILE_PAGE_EXTRACT)
+    assert _get_total_members_from_profile(profile_soup) == 42